Real world hypoglycaemia related to glucose variability and Flash glucose scan frequency assessed from global FreeStyle Libre data

Abstract

Aim: Flash glucose monitoring provides a range of glucose metrics. In the current study, we aim to identify those that indicate that glycaemic targets can be consistently met and contrast the total (t-CV) and within-day coefficient of variation (wd-CV) to guide the assessment of glucose variability and hypoglycaemia exposure.Methods: De-identified data from Flash readers were collected. The readers were sorted into 10 equally sized groups of scan frequency followed by quartiles of estimated A1c (eA1c). A similar grouping was performed for the total coefficient of variation (t-CV) and within-day coefficient of variation (wd-CV). In addition, analysis of the association of time below 54 mg/dl and glucose variability measured by t-CV and wd-CV was performed. Results:The dataset included 1 002 946 readers. Readers sorted by 10 equal groups of scan rate and quartiles by eA1c, t-CV and wd-CV represented 25 074 readers per group. The association of lower eA1c with higher time in range and reduced time above range was clear. The correlation of eA1c quartiles and time below range was not consistent. An association between glucose variability and hypoglycaemia was found. Both wd-CV and t-CV were associated with time below range. For achieving the consensus target of <1% time below 54 mg/dl, the associated wd-CV and t-CV values were 33.5% and 39.5%, respectively. Conclusions:The type of CV reported by the different continuous glucose monitoring systems should be acknowledged. CV <36% might not be adequate to ensure low hypoglycaemia exposure. To our knowledge, the majority of continuous glucose monitoring reports the t-CV. Appropriate thresholds should be used to identify patients that would probably meet time below range targets (t-CV <40% or wd-CV <34%).

The significance of coefficient of variation as a measure of hypoglycaemia risk and glycaemic control in real world users of the automated insulin delivery MiniMed 780G system

“…Other recent publications confirm that CV thresholds for hypoglycaemia prediction deviate from the historic 36% threshold value. Riveline et al 11 analysed data from a large intermittently scanned CGM database and noted that the CV thresholds for hypoglycaemia prediction were related to the method of calculation, specifically within day CV versus total CV (equivalent to the CV metric used in this study). When using total CV, they found a threshold of total CV <39.5% and suggested that CV <36% may not be adequate to ensure low hypoglycaemia exposure.…”

“…When using total CV, they found a threshold of total CV <39.5% and suggested that CV <36% may not be adequate to ensure low hypoglycaemia exposure. The difference between the studies is probably related to the treatment modalities, as in Riveline et al 11 the therapies were based on non‐automated insulin systems, while the present analysis utilized data from people with type 1 diabetes treated with the MM780G system. It is possible that the findings of both analyses may be influenced by parameters such as time spent wearing the sensor; the findings of Riveline et al, however, could also potentially be influenced by other parameters such as number of scans per day.…”

“…The validation set comprised the remaining 3026 users in the total analysis set, which was then used to assess the rate of correct classification with the CV cut‐off point found in the test set. The rate of correct classification with the CV cut‐off point that was identified in the present analysis was then compared with the rate of correct classification with CV <36% and CV <39.5%, as suggested by Monnier et al 7 and Riveline et al, 11 respectively.…”

“…The reference threshold of CV <36% was selected from the upper limit of the distribution of CVs in a reference group of 82 people with type 2 diabetes with mean age of 63 years on either dietary measures alone (n = 8) or treatment combining diet with insulin sensitizers (n = 74) 7 . However, based on intermittently scanned CGM readings, Riveline et al have recently suggested a CV threshold of 39.5%, which challenges the use of the current consensus threshold value 11 . We also argue that the use of 36% as the CV of the SG threshold in the treatment of people with type 1 diabetes using AID with a system such as the MiniMed 780G (MM780G) system is questionable and requires a thorough evaluation.…”

“…7 However, based on intermittently scanned CGM readings, Riveline et al have recently suggested a CV threshold of 39.5%, which challenges the use of the current consensus threshold value. 11 We also argue that the use of 36% as the CV of the SG threshold in the treatment of people with type 1 diabetes using AID with a system such as the MiniMed 780G (MM780G) system is questionable and requires a thorough evaluation.…”